(1) Field of the Invention
The present invention relates to end-point detection in a processing on a to-be-processed layer using plasma etching or the like.
(2) Description of Related Art
With the increasing miniaturization of semiconductor devices, gate oxide films have become thinner. As a result, dry etching of a polysilicon film that will be a gate electrode has required detection of an end point of etching before exposure of a gate oxide film. To meet the requirement, a method for detecting an end point using interference lights obtained by the interference of reflected lights, as disclosed for example in Japanese Unexamined Patent Publication No. 2001-85388, allows the end point of etching to be detected before exposure of a gate oxide film.
First, an etching device to which a known end-point detection method is applied will be described with reference to FIG. 10.
An etching device shown in FIG. 10 comprises a process chamber 10, a lower electrode 12 positioned on the bottom of the process chamber 10 and serving as a sample stage on which a semiconductor substrate 11 is to be put, and a top plate 13 placed above and located a predetermined distance apart from the lower electrode 12. A coil 14 is placed on the top plate 13. The process chamber 10 is provided with a gas supply part (not shown) and a gas-emitting part 15 to which an evacuator (not shown) is connected. RF power sources 16 are connected via matching boxes (not shown) to the lower electrode 12 and a coil 14, respectively. Power is applied from the RF power sources 16 to the lower electrode 12 and the coil 14, thereby etching the semiconductor substrate 11.
As further shown in FIG. 10, a window member 17 is mounted to the top plate 13, and a light-receiving/emitting device 19 is placed above the window member 17. The light-receiving/emitting device 19 irradiates the semiconductor substrate 11 with light from a light source 18 and collects lights reflected on the semiconductor substrate 11. The light source 18 is connected via an optical fiber 20 to the device 19, the device 19 is connected via an optical fiber 20 to a spectroscope 21, and the spectroscope 21 is in turn connected via an optical fiber 20 to an end-point detection apparatus 22, respectively. Furthermore, the end-point detection apparatus 22 is connected to a control section 23 of the etching device. When an end-point detection signal 24 is transmitted from the end-point detection apparatus 22 to the control section 23, the control section 23 transmits, to the RF power sources 16, signals 25 for stopping the application of power in order to complete etching.
Next, the known end-point detection method will be described with reference to FIGS. 10, 11A and 11B.
As shown in FIGS. 10 and 11A, light from a light source (light source 18) is radiated vertically to the surface of a Si substrate (semiconductor substrate 11) In this relation, a polysilicon film is formed above the Si substrate with a gate oxide film interposed therebetween. While part of the radiated light from the light source is reflected on the polysilicon film, another part of the radiated light transmits the polysilicon film and is then reflected on the interface between the gate oxide film and the polysilicon film. These reflected lights interfere with each other to form interference light. The interference light is recaptured through the window member 17 of the top plate 13, the device 19, the optical fiber 20, and the spectroscope 21 by the end-point detection apparatus 22. In this case, the waveform of the interference light (time variations in the intensity of the interference light) varies depending on the thickness of the remaining polysilicon film. Thus, the monitoring of the interference light waveform enables the detection of the instant at which the thickness of the remaining polysilicon film reaches a desired value, i.e., an end point of etching.
FIG. 11A shows the paths of radiated light beams from the light source and the paths of the reflected light beams from the polysilicon film and the like. In this figure, these paths are tilted for convenience.
A description will be given below of the relationship between the thickness d of the remaining polysilicon film and interference light during etching with reference to FIGS. 11A and 11B.
As shown in FIG. 11A, there is a path difference 2d between the reflected light from the polysilicon film and the reflected light from the interface between the gate oxide film and the polysilicon film. Therefore, when the light from the light source is radiated vertically to the substrate, this path difference 2d wholly constitutes a phase difference between the two reflected lights to produce interference light. Accordingly, as shown in FIG. 11B, when this phase difference is any integral multiple of the wavelength of the reflected light in the polysilicon film, the intensity of the interference light becomes strongest. On the other hand, when this phase difference is shifted a half wavelength from any integral multiple of the wavelength of the reflected light in the polysilicon film, the intensity of the interference light becomes weakest. More particularly, with the increase in etching amount, the intensity of the interference light periodically varies.
In this case, the relationship between the thickness d of the remaining polysilicon film and the interference light is represented by the following equations (1) through (3):Intensity of Interference Light=A2+B2+2AB×cos(a−b)  (1)Phase Difference (a−b)=2πn×(2d/λ)  (2)Remaining Film Thickness d=d0−Rt  (3)wherein A and B represent the respective amplitudes of the reflected lights, a and b represent the respective initial phases of the reflected lights, n represents an integer, d represents the thickness of the remaining polysilicon film, d0 represents the initial thickness of the polysilicon film, R represents the etching rate of the polysilicon film, t represents etching time, and λ represents the wavelength of the light.
A description will be given below of a known method for fabricating a semiconductor device and more particularly of a known method for forming a gate electrode with reference to cross-sectional views showing process steps in FIGS. 12A through 12C.
First, as shown in FIG. 12A, a gate oxide film 33 is formed by thermal oxidation or the like on a semiconductor substrate 31 formed with an isolation region 32. Then, a polysilicon film 34 and a silicon oxide film 35 are successively formed thereon by a film forming method such as Chemical Vapor Deposition (CVD). Thereafter, a desired gate resist pattern 36 is formed thereon using photolithography.
Next, as shown in FIG. 12B, dry etching is performed on the silicon oxide film 35 by a dry etching technique using the gate resist pattern 36 as a mask. Thereafter, washing and cleaning are carried out to remove the gate resist pattern 36. In this way, a silicon oxide film 35A is formed, to which a gate pattern has been transferred.
Next, as shown in FIG. 12C, dry etching is performed on the polysilicon film 34 by pattern has been transferred. FIG. 12C shows the state of the semiconductor device immediately after detection of an end point of etching.
In the dry etching of the polysilicon film 34 shown in FIG. 12C, the end point of etching is detected by the above-mentioned known end-point detection method using interference light. That is, the end point of etching is detected on the basis of a desired value of the thickness of the remaining polysilicon film 34. More particularly, the desired value of the thickness of the remaining film is 50 nm, the wavelength of the light is 600 nm, and the instant at which the waveform of the interference light reaches the second maximum is detected as the end point of etching. The reason why the maximum of the waveform of the interference light is used to detect the end point of etching is that human beings can easily judge whether the end point of etching is normally detected. Note that, as represented by the equations (1) through (3), the end point of etching can be detected without the use of the maximum or the minimum, and the wavelength of the light can be set at an arbitrary value.
However, according to the above-mentioned known end-point detection method used for gate formation, in dry etching of the polysilicon film 34, the waveform of the interference light obtained during the detection of an end point is not sinusoidal but forms, for example, a waveform as shown in FIG. 13. As a result, as shown in FIG. 12D, it becomes impossible to detect the end point, resulting in the damaged gate oxide film 33. More particularly, for example, through holes 37 are produced in the gate oxide film 33.
On the other hand, in the end-point detection method using interference light as disclosed in Japanese Unexamined Patent Publication No. 2001-85388, in order to cancel the influence of reflected light from the resist film serving as a mask, a dummy end point is detected using two kinds of reflected lights of different wavelengths. The two kinds of reflected lights have a relationship in which the phase differences of interference waveforms each having a short period is close to each other. To be specific, this method determines the intensity ratio between respective interference lights from the two kinds of reflected lights or a differential value of the intensity ratio. With this method, the phase difference between respective interference waveforms from the two kinds of reflected lights is determined on the basis of the intensity ratio between the two interference waveforms. Thus, detected as a dummy end point is the instant at which the intensity ratio (=phase difference) approaches a constant value or the instant at which the differential value of the intensity ratio (=phase difference) approaches 0. For example, as shown in FIG. 12C, the instant at which the polysilicon film 34 is slightly left is set as the dummy end point using a sample wafer in order to prevent the gate oxide film 33 from being overetched. Therefore, product wafers of different wafer structures (for example, mask layouts) have somewhat different interference waveforms as compared with the sample wafer due to the influence of the aperture ratio of the resist film, pattern dependence, or variations in dimensions in a lithography process. As a result, the thickness of the remaining polysilicon film to be etched cannot precisely be controlled. Hence, the end-point detection time at which the end point is detected varies depending on the wafer structures, resulting in the damaged gate oxide film. Furthermore, after the detection of the end point, the etching conditions are switched to high-selectivity conditions, and controllability over dimensions is degraded under high-selectivity conditions. Therefore, when the end-point detection time varies depending on the wafer structures, the gate length varies from one wafer to another.